Title: Predicting Highway Safety for Curves on Two-Lane Rural Highway
1HSM Practitioners Guide for Two-Lane Rural
Highways Workshop
- Predicting Highway Safety for Curves on Two-Lane
Rural Highway
- Session 4
2Predicting Highway Safety for Curves on Two-Lane
Rural Highways
- Describe the crash prediction method for Crash
Performance on Horizontal Curves - Identify low-cost safety improvements for
horizontal curves
3.Curves present particular safety problems to
designers
CRASH RATES (Crashes per 1 km segment--3 year
timeframe)
- The risk of a reported crash is about three
times greater on a curve than on a tangent
Crash Rate
Source Glennon, et al, 1985 study for FHWA
Tangent segments
Segments w/curve
Curved portion only (Curve plus transitions)
4(No Transcript)
5(No Transcript)
6(No Transcript)
7(No Transcript)
8(No Transcript)
9Actual Driver Operations on Curves
Driver tracks a critical radius sharper than
that of the curve just past the PC
- Drivers overshoot the curve (track a path
sharper than the radius) - Path is a spiral
- Path overshoot behavior is independent of speed
Source Bonneson, NCHRP 439 and Glennon et al
(FHWA)
10Driver overshoot behavior on curves (from
Glennon, et al)
700
Example -- a 1000-ft radius curve is driven by a
95th percentile driver at about a 700 ft
radius at some point in the curve
11Research confirms differences in actual
operations versus AASHTO assumptions
- Drivers selected speed behavior does not match
design assumptions - Sharper curves (lt80 km/h or 50 mph) are driven
faster (drivers are more comfortable)
Curves driven faster than Policy assumption
Curves driven slower than Policy assumption
12Speed Prediction Model for Horizontal Curves
(Otteson and Krammes)
V85 41.62 - 1.29D 0.0049L - 0.12DL 0.95 Vt
- Where
- V85 85th percentile speed on the curve
- D degree of curve
- L length of curve (mi)
- Vt 85th percentile approach speed (mph)
- this should be measured in the field
13A risk assessment tool for speed profiles
- V85 - Vdesign Vdelta
- Higher risk curves may be those with V delta
high (i.e., operating speeds significantly
greater than design speed) - Vdelta gt 12 mph (20 km/h) high risk
- 6 mph (10 km/h) lt Vdelta lt 12 mph (20 km/h)
caution
14FHWAs IHSDM Speed Consistency Model Addresses
Continuous Speed Behavior
15Truck operations on curves may in some cases be
critical (Harwood and Mason)
- Under certain conditions, trucks will roll over
before they skid - Trucks with high centers of gravity overturn
before losing control due to skidding - Margin of safety for f is therefore lower for
trucks - Trucks on downgrade curves generate greater
lateral friction (superelevation is not as
effective)
16Summary of Research on Superelevation and
Transition Design
- Studies confirm small but significant effect of
superelevation on crashes - FHWA (Zegeer) study noted 5 to 10 greater
crashes when superelevation is deficient - 1987 study of fatal crash sites on curves noted
deficiencies in available superelevation
17Research confirms benefits of spirals and
recommends optimal transition design
Spirals provides e transition leading into the
curve
Radius (m)
Source NCHRP Report 439
- Zegeer et al found safety benefits in HSIS study
of Washington - Bonneson confirmed operational benefits noted by
Glennon, etal
18Zegeer et al. FHWA Study Cost-Effective
Geometric Improvements for Safety Upgrading of
Horizontal Curves (1991)
- Data Bases
- 10,900 Curves in Washington State
- 7-state data base of 5000 mi
- 78 curves in New York State
- Glennon 4-state data base of 3277 curve segments
- Statistical Analysis and Model Development
- Identified as key effort in TRB SR 214, recent
NCHRP review by BMI, and key reference for IHSDM
19Summary of findings from Zegeer study
- Features related to crashes include
- Degree and length of curve
- Width through the curve
- Superelevation and,
- Spiral presence
- For typical volumes on 2-lane highways, expect 1
to 3 crashes per 5 years on a curve
20Safety Effects for Horizontal Curves (CMF3r)
- CMF3r 1.55 Lc (80.2/R) - 0.012 S
1.55Lc - Where
- Lc Length of Curve including spirals, (mi)
- R Radius of Curve (ft)
- S 1 if spiral transition is present, 0 if not
present
21Safety Effects of Horizontal Curves (CMF3r)
Example with no Spiral present
- For Lc 480 feet 0.091 miles
- R 350 no spiral transition
CMF3r 1.55 Lc (80.2/R) 0.012S / 1.55Lc
(1.55 x 0.091) (80.2/350)
0.012x0 1.55x 0.091
2.62
22Safety Effects of Horizontal Curves (CMF3r)
Example with Spiral Transition
- For Lc 480 feet 0.091 miles
- R 350 with spiral transition
CMF3r 1.55 Lc (80.2/R) 0.012S / 1.55Lc
(1.55 x 0.091) (80.2/350)
0.012x1 1.55x 0.091
?
2.54
Without spiral CMF3r 2.62, with spiral CMF3r
2.54, Difference 8 potential for fewer
crashes with a spiral transition in this segment.
23Crash Modification Function for Horizontal
Curves Superelevation
CMF4r is based on Superelevation variance or SV
- For SV less than 0.01 CMF4r 1.00
- For 0.01 lt SV lt 0.02 CMF4r 1.00 6(SV-0.01)
- For SV gt 0.02 CMF4r 1.06 3(SV-0.02)
Example Design e 4, Actual e 2
SV 0.04 0.02 0.02
CMF4r 1.06 3(0.02-0.02) 1.06 3(0.0) 1.06
24HSM Applications to Two-Lane Rural Highway
Segments
- HSM Crash Prediction Method for Two-Lane Rural
Highway Segments - Applying SPF and CMFs
- Example Problem
25Crash Prediction for Roadway Segment for Existing
Conditions Example Calculation
- Two-Lane Rural Roadway, CR 123 Anywhere, USA (MP
10.00 15.02) - AADT 3,500 vpd for the current year
- Length 26,485 feet 5.02 miles
- Lane Width 11.0 ft
- Shoulder Width 2 ft Shoulder Type Gravel
- Horizontal Curve on Grade (MP 12.00-12.186)
- Lc 0.186 miles, R 650 with no spiral
transition - Grade 4.5
- Superelevation Variance .02
- Tangent Section on Grade (MP 13.45-14.00)
- L 0.55 miles Grade -6.3
26Crash Prediction for Roadway Segment for Existing
Conditions Example
- Divide Two-Lane Rural Roadway into Individual
Segments
Segment Length (miles) Horizontal Curve Radius (ft) Super-elevation Variance Grade () Driveway Density (per mile) RHR
10.00 12.00 2.000 Tangent N/A 2.0 8 5
12.00 12.186 0.186 650 .02 4.5 0 5
12.186 -13.45 1.264 Tangent N/A 3.0 4 5
13.45-14.00 0.550 Tangent N/A - 6.3 0 5
1400-15.02 1.020 Tangent N/A - 3.0 6 5
27Safety Performance Function (SPF) for Base
Conditions Example Calculation
Segment 2 (MP 12.00-12.186) Horizontal Curve on
a 4.5 Grade
- Where
- AADT 3,500 vpd (current year)
- Length 0.186 miles
Nspf-rs (AADTn) (L) (365) (10-6) e-0.312
Nspf-rs (3,500) (0.186) (365) (10-6) e-0.312
(3,500) (0.186) (365) (10-6) (0.7320)
0.17 crashes per year
28CMF for Lane Width (CMF1r) Calculation
Segment 2 11 foot wide lane
From Table 10-8 CMFra 1.05
- Adjustment for lane width and shoulder width
related crashes (Run off Road Head-on
Sideswipes) to obtain total crashes using default
value for pra 0.574
CMF1r (CMFra - 1.0) pra 1.0
(1.05 - 1.0) 0.574 1.0
(0.05) (0.574) 1.0
1.03
29CMF or Shoulder Width and Type (CMF2r)
Calculation
Segment 2 2 ft wide gravel shoulder
CMFwra 1.30 (Table10-9) and CMFtra 1.01
(Table10-10)
- Adjustment from crashes related to lane and
shoulder width (Run off Road Head-on
Sideswipes) to total crashes using default value
for pra 0.574
CMF2r (CMFwra CMFtra - 1.0) pra 1.0
((1.30)(1.01) - 1.0) 0.574 1.0
(0.313) (0.574) 1.0
1.18
30CMF for Horizontal Curve (CMF3r) Calculation
Segment 2 Horizontal Curve
- For Lc 0.186 miles
- R 650 with no spiral transition
CMF3r 1.55 Lc (80.2/R) 0.012S / 1.55Lc
(1.55 x 0.186) (80.2/650)
0.012x0 1.55x
0.186
1.43
31CMF for Superelevation on Horizontal Curves
(CMF4r)
Segment 2 Horizontal Curve Superelevation
Variance 0.02
- For SV gt 0.02 CMF4r 1.06 3(SV-0.02)
CMF4r 1.06 3(0.02-0.02) 1.06 3(0.0)
1.06
32CMF for Percent () Grade on Roadway Segments
(CMF5r)
Segment 2 4.5 Grade
CMF5r 1.10
33CMF Roadside Design (CMF10r) Example Calculation
CMF10r e(-0.6869 (0.0668xRHR)) /e-0.4865
e(-0.6869 (0.0668x5)) /e-0.4865
1.14
34Applying CMFs to the SPF Base Prediction Model
CRASH MODIFCATION FACTORS Lane
Width 11 ft CMF1r 1.03 Shoulder Width 2 ft
gravel CMF2r 1.18 Horizontal Curve CMF3r
1.43 Superelevation Variance (0.02) CMF4r
1.06 Percent Grade 4.5 CMF5r 1.10 Driveway
Density, None CMF6r 1.00 Centerline Rumble,
None CMF7r 1.00 Passing/Climbing Lanes,
None CMF8r 1.00 TWLTLs, None CMF9r
1.00 Roadside Design, RHR 5 CMF10r
1.14 Lighting, None CMF11r 1.00 Automated
Enforcement, None CMF12r 1.00
Segment 2 SPF and CMF Values AADT 3,500
vpd, Length 0.186 mi Radius 650
ft Nspf-rs 0.17 crashes per year CMFtotal 2.31
35Applying CMFs to the SPF Base Prediction Model
Npredicted-rs Nspf-rs x (CMF1r CMF12r) Cr
Segment 2 Apply CMFs to SPF for Base Conditions
(letting Cr 1.0)
0.17 x (1.03 x 1.18 x 1.43 x 1.06 x 1.10 x 1.00 x
1.00 x 1.00 x 1.00 x 1.14 x 1.000 x 1.00) x 1.00
Npredicted-rs
0.17 x 2.31 x 1.00
0.4 crashes per year, 1 crash every 2.5 yrs
36Crash Prediction for Roadway Segment for Existing
Conditions Example Calculation
- For each Two-Lane Rural Roadway Segment Table
with SPF predicted crahses, CMFs, and Adjusted
Total Crashes
CRASH PREDICTION METHOD TOTAL CRASHES CRASH PREDICTION METHOD TOTAL CRASHES CRASH PREDICTION METHOD TOTAL CRASHES CRASH PREDICTION METHOD TOTAL CRASHES CRASH PREDICTION METHOD TOTAL CRASHES CRASH PREDICTION METHOD TOTAL CRASHES CRASH PREDICTION METHOD TOTAL CRASHES CRASH PREDICTION METHOD TOTAL CRASHES CRASH PREDICTION METHOD TOTAL CRASHES CRASH PREDICTION METHOD TOTAL CRASHES CRASH PREDICTION METHOD TOTAL CRASHES CRASH PREDICTION METHOD TOTAL CRASHES CRASH PREDICTION METHOD TOTAL CRASHES CRASH PREDICTION METHOD TOTAL CRASHES CRASH PREDICTION METHOD TOTAL CRASHES
Seg No. SPF base CMF1r LW CMF2r SWST CMF3r ST CMF4r e CMF5r Grade CMF6r DD CMF7r CLRS CMF8r PassLn CMF9r TWLTL CMF10r RD CMF11r Light CMF12r Spd Enf Total CMF Total Adjusted Crashes
1 1.87 1.03 1.18 1.00 1.00 1.00 1.07 1.00 1.00 1.00 1.14 1.00 1.00 1.49 2.8
2 0.17 1.03 1.18 1.43 1.06 1.10 1.00 1.00 1.00 1.00 1.14 1.00 1.00 2.31 0.40
3 1.27 1.03 1.18 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.14 1.00 1.00 1.39 1.8
4 0.51 1.03 1.18 1.00 1.00 1.16 1.00 1.00 1.00 1.00 1.14 1.00 1.00 1.61 0.8
5 0.95 1.03 1.18 1.00 1.00 1.00 1.02 1.00 1.00 1.00 1.14 1.00 1.00 1.42 1.4
Total 7.2
37Predicting Crash Frequency Performance
Total Predicted Crash Frequency within the limits
of the roadway being analyzed
Ntotal crashes ?Npredicted-rs ? Npredicted-int
Ntotal crashes 7.2 crashes/yr ? Npredicted-int
38Overview of Good Alignment Design Practice
(suggested by safety and operational research)
- Curves and grades are necessary features of
alignment design (reflect the topography,
terrain, and context) - Pay particular attention to roadside design
adjacent to curves - Avoid long, sharp curves
- Adjust alignment design to reflect expected
speeds on curves
39Overview of Good Alignment Design Practice
(continued)
- Avoid minimum radius designs where
- actual speeds will be higher than design speeds
- truck volumes will be substantial
- combined with steep grades
- Use spiral transition curves, particularly for
higher speed roads and sharper curves
40Overview of Good Alignment Design Practice
(continued)
- Minimize grades within terrain context
- Widen lanes and shoulders through curves
- Pay attention to access points related to
horizontal and vertical curve locations
41Low and Lower Cost Safety Improvements for
Horizontal Curves
42Low Cost Intersection Safety Measures Signing
Countermeasures
Advance Warning With Speed Advisory
Injury Crashes CMF 0.87 CRF 13
PDO Crashes CMF 0.71 CRF 29
CMF Clearinghouse http//www.cmfclearinghouse.org
43Safety Effects of Installing Combination
Horizontal Alignment Warning Advisory Speed
Signs
44Signing Countermeasure for Horizontal Curves
Chevrons Signs
CRF 35
CMF 0.65
CMF Clearinghouse http//www.cmfclearinghouse.org
45Safety Effects of Installing RPMs
46Low Cost Intersection Safety Measures Signing
Countermeasures
Double Up Advance Warning Signs
CRF 31 CMF 0.69
47Low Cost Intersection Safety Measures Signing
Countermeasures
Sharp 10 mph curve to right just over hill
Activated Warning Beacon
- Radar activated flasher when speed is fast for
10mph curve
48Examples of Improving Safety of Existing Curves
Widen Shoulders
- Widen 2 Shoulder to 6 Shoulder NY Rte 82
north of Millbrook
6
2
49Examples of Improving Safety of Existing Curves
Widen Shoulder on Inside of Tight Curve
- Widening on
- Inside of Curves
NCHRP 500, Strategy 15.2 A11 Widening in Curves
50Low Cost Intersection Safety Measures Signing
Countermeasures
- 9. Illumination of
- Rural Curves
CRF 28 for injury crashes highway lighting
Route 376 near Poughkeepsie, NY
51Predicting Highway Safety for Curves on Two-Lane
Rural Highways
- Described the equation for prediction of Crash
Performance on Horizontal Curves - Identified low-cost safety improvements for
horizontal curves
52Questions and Discussion